DeFFusion: CNN-based Continuous Authentication Using Deep Feature Fusion

Smartphones have become crucial and important in our daily life, but the security and privacy issues have been major concerns of smartphone users. In this article, we present DeFFusion, a CNN-based continuous authentication system using Deep Feature Fusion for smartphone users by leveraging the accelerometer and gyroscope ubiquitously built into smartphones. With the collected data, DeFFusion first converts the time domain data into frequency domain data using the fast Fourier transform and then inputs both of them into a designed CNN, respectively. With the CNN-extracted features, DeFFusion conducts the feature selection utilizing factor analysis and exploits balanced feature concatenation to fuse these deep features. Based on the one-class SVM classifier, DeFFusion authenticates current users as a legitimate user or an impostor. We evaluate the authentication performance of DeFFusion in terms of impact of training data size and time window size, accuracy comparison on different features over different classifiers and on different classifiers with the same CNN-extracted features, accuracy on unseen users, time efficiency, and comparison with representative authentication methods. The experimental results demonstrate that DeFFusion has the best accuracy by achieving the mean equal error rate of 1.00% in a 5-second time window size.

Blindfold: A Smartphone based Object Detection Application

With the advancement of computing power of Smartphones, they seem to be a better option to be used as an Assistive Technology for the visually impaired. In this paper we have discussed an application which allows visually impaired users to detect objects of their choice in their environment. We have made use of the Tensorflow Lite Application Programmable Interface (API), an API by Tensorflow which specifically runs models on an Android Smartphone. We have discussed the architecture of the API and the application itself. We have discussed the performance of various types of models such as MobileNet, ResNet & Inception. We have compared the results of the various Models on their size, accuracy & inference time(ms) and found that the MobileNet has the best performance. We have also explained the working of our application in detail.

Optimizing Costs to Collect Local Infauna through Grabs: Effect of Sampling Size and Replication

Most ecological studies require a cost-effective collection of multi-species samples. A literature review unravelled that (1) large-sized grabs to collect infauna have been used at greater depths, despite no consistent relationship between grab size and replication across studies; and (2) the total number of taxa and individuals is largely determined by the replication. Then, infauna from a sedimentary (sandy) seabed at Gran Canaria Island was collected through van Veen grabs of three sizes: 0.018, 0.042 and 0.087 m2 to optimize, on a simple cost-benefit basis, sample size and replication. Specifically, (1) the degree of representativeness in the composition of assemblages, and (2) accuracy of three univariate metrics (species richness, total infaunal abundances and the Shannon-Wiener index), was compared according to replication. Then, by considering mean times (a surrogate of costs) to process a sample by each grab, (3) their cost-efficiency was estimated. Representativeness increased with grab size. Irrespective of the grab size, accuracy of univariate metrics considerably increased when n > 10 replicates. Costs associated with the 0.087 m2 grab were consistently lower than costs by the other grabs. In conclusion, because of high representativeness and low cost, a 6.87 L grab appears to be the optimal sample size to assess infauna at our local site.

Determination of the minimum number of technological bases of a part

The selection of technological bases is considered before the choice of the type of billet and the development of the route of the technological process. A technique is proposed for selecting the minimum number of sets of technological bases according to the criterion of equality in the cost price of manufacturing the part according to the principle of unity and combination of bases at this stage. Keywords: part, surface, coordinating size, accuracy, design and technological base, labor input, cost price. [email protected]

Estimating the proportion of overweight soldiers in the Australian Army by combat uniform waist size

IntroductionRecent studies have shown an increasing number of overweight and obese members serving in many armies. Overweight and obesity can be estimated using either body mass index or waist circumference measures. The aim of this research was to estimate the proportion of the Australian Army considered to be overweight and obese by waist circumference using the proxy measure of issued combat uniform waist size.MethodThe Australian Army has been progressively replacing combat uniforms with a new uniform design and camouflage pattern since 2016. The total number of issued combat uniforms by size was obtained from the points of issue for the three Australian Army combat brigades from the first issue of the new uniform in January 2016 through to November 2019. The waist size of issued combat pants was collated from each of the three points of issue, adjusted for measured waist size accuracy and sex, and analysed to estimate the proportion of overweight and obese soldiers in the Australian Army.ResultsThere were a total of 155 735 combat pants issued across the three points of issue. The mean waist size based on combat uniform pant size was found to be 90.4 cm, with an SD of 7.5 cm. Based on these data, approximately 23.3% of the Australian Army population can be estimated to be overweight and an additional 4.5% to be obese.ConclusionsThe Australian Army, like many western armies, has a significant proportion of overweight personnel. This can negatively affect operational capability, health and future healthcare costs both within the military and to society after military service has concluded. This is the first study to use a uniform waist size as a proxy to estimate overweight and obesity. This technique has application for the military, emergency services or any other organisation in which uniforms are provided.

Overview of Recent Advances in the Design of Plasmonic Fiber-Optic Biosensors

Plasmonic fiber-optic biosensors combine the flexibility and compactness of optical fibers and high sensitivity of nanomaterials to their surrounding medium, to detect biological species such as cells, proteins, and DNA. Due to their small size, accuracy, low cost, and possibility of remote and distributed sensing, plasmonic fiber-optic biosensors are promising alternatives to traditional methods for biomolecule detection, and can result in significant advances in clinical diagnostics, drug discovery, food process control, disease, and environmental monitoring. In this review article, we overview the key plasmonic fiber-optic biosensing design concepts, including geometries based on conventional optical fibers like unclad, side-polished, tapered, and U-shaped fiber designs, and geometries based on specialty optical fibers, such as photonic crystal fibers and tilted fiber Bragg gratings. The review will be of benefit to both engineers in the field of optical fiber technology and scientists in the fields of biosensing.

Exploring compression and parallelization techniques for distribution of deep neural networks over Edge–Fog continuum – a review

PurposeThe trend of “Deep Learning for Internet of Things (IoT)” has gained fresh momentum with enormous upcoming applications employing these models as their processing engine and Cloud as their resource giant. But this picture leads to underutilization of ever-increasing device pool of IoT that has already passed 15 billion mark in 2015. Thus, it is high time to explore a different approach to tackle this issue, keeping in view the characteristics and needs of the two fields. Processing at the Edge can boost applications with real-time deadlines while complementing security.Design/methodology/approachThis review paper contributes towards three cardinal directions of research in the field of DL for IoT. The first section covers the categories of IoT devices and how Fog can aid in overcoming the underutilization of millions of devices, forming the realm of the things for IoT. The second direction handles the issue of immense computational requirements of DL models by uncovering specific compression techniques. An appropriate combination of these techniques, including regularization, quantization, and pruning, can aid in building an effective compression pipeline for establishing DL models for IoT use-cases. The third direction incorporates both these views and introduces a novel approach of parallelization for setting up a distributed systems view of DL for IoT.FindingsDL models are growing deeper with every passing year. Well-coordinated distributed execution of such models using Fog displays a promising future for the IoT application realm. It is realized that a vertically partitioned compressed deep model can handle the trade-off between size, accuracy, communication overhead, bandwidth utilization, and latency but at the expense of an additionally considerable memory footprint. To reduce the memory budget, we propose to exploit Hashed Nets as potentially favorable candidates for distributed frameworks. However, the critical point between accuracy and size for such models needs further investigation.Originality/valueTo the best of our knowledge, no study has explored the inherent parallelism in deep neural network architectures for their efficient distribution over the Edge-Fog continuum. Besides covering techniques and frameworks that have tried to bring inference to the Edge, the review uncovers significant issues and possible future directions for endorsing deep models as processing engines for real-time IoT. The study is directed to both researchers and industrialists to take on various applications to the Edge for better user experience.

Probabilistic Neural Network and Genetic Algorithm for Abdominal Aorta Aneurysm Identification

An abdominal aorta aneurysm (AAA) can cause severe threat if it burst. Doctors can detect the presence of AAA by using abdominal ultrasound. As the treatment depends on the location and size, accuracy plays a significant role. To prevent devastating clinical outcome in this proposed work, new approaches and algorithms were used for generating the infallible result. After processing the AAA image by using notch filter, exudate based segmentation is performed and the selected features gets classified by using probabilistic neural network classifier. By using PNN classifier, accuracy and sensitivity gets enhanced in this work. The achieved accuracy is 98% and sensitivity 97.5%. While analogizing the proposed work with other existing work. It’s very facile to perform and expected target gets achieved

Eyes on the size: accuracy of visual length estimates of white sharks, Carcharodon carcharias

Visual estimates have been used extensively to determine the length of large organisms that are logistically challenging to measure. However, there has been little effort to quantify the accuracy or validity of this technique despite inaccurate size estimates leading to incorrect population assessments and misinformed management strategies. Here, we compared visually estimated total length measurements of white sharks, Carcharodon carcharias , during cage-diving operations with measurements obtained from stereo-video cameras and assessed the accuracy of those estimates in relation to suspected biases (shark size, and observer experience and gender) using generalized linear mixed-models and linear regressions. Observer experience on board cage-diving vessels had the greatest effect on the accuracy of visual length estimates, with scientists being more accurate (mean accuracy ± standard error: 23.0 ± 16.5 cm) than crew (39.9 ± 33.8 cm) and passengers (49.4 ± 38.5 cm). Observer gender and shark size had no impact on the overall accuracy of visual length estimates, but passengers overestimated sharks less than 3 m and underestimated sharks greater than 3 m. Our findings show that experience measuring animals is the most substantial driver of accurate visual length estimates regardless of the amount of exposure to the species being measured. Scientists were most accurate, even though crew observe white sharks more frequently. Our results show that visual length estimates are not impacted by shark size and are a valid measurement tool for many aspects of C. carcharias research, provided they come from people who have previously been involved in measuring animals, i.e. scientists .